Get MIN MAX grouping by closer distance - sql

Question

shops
id_shop | id_prod_ty | position
and
products
id | price | id_prod_ty | distance
table shops contains 2 shops for each id_prod_ty (product type) and a position.
table products contains many records with different prices and a distance
basically i need to have a query that selects price low and high for each product and for each shops grouping by closer prices (and that are not closer to other shops)
so for example
shops
id_shop | id_prod_ty | position
1 | 1 | 3
2 | 1 | 7
3 | 2 | 8
4 | 2 | 4
....
products
id | price | id_prod_ty | distance
1 | 10 | 1 | 1
2 | 04 | 1 | 2
3 | 02 | 1 | 4
4 | 44 | 1 | 2
5 | 09 | 1 | 1
6 | 13 | 1 | 7
7 | 15 | 1 | 8
8 | 09 | 2 | 5
9 | 12 | 2 | 8
10 | 17 | 2 | 1
11 | 32 | 2 | 13
12 | 22 | 2 | 2
...
result shout be like this
id_prod_ty | id_price_low | id_price_high | id_shop
1 | 3 (02) | 4 (44) | 1
1 | 6 (13) | 7 (15) | 2
2 | 8 (09) | 12 (22) | 4
2 | 9 (12) | 11 (32) | 3
...
thanks

Answer 1

I think I understand the problem. For each price, you are trying to assign the shop that is nearest, based on the distance and the position.
The approach starts by joining the positions and shops. It then calculates the difference between the position and distance -- that seems to be the measure you are using. Each price id will appear twice (once for each shop). Using a window function, it calculates the minimum difference for each id.
With this information, the query can determine which is the closer shop. This version returns the prices, rather than the ids of the lowest prices. Here is the final, untested, query:
select id_prod_ty,
MIN(case when diff = minDiff then price) as price_low,
MAX(case when diff = minDiff then price) as price_high,
s.id_shop
from (select p.id_prod_ty,
s.id_shop,
p.price,
(p.distance - s.position) as diff,
MIN(p.distance - s.position) over (partition by p.id) as minDiff
from products p join
shops s
on p.id_prod_ty = s.id_prod_ty
) ps
group by id_prod_ty, id_shop
Here is a variation that gets the ids as well. It uses a window function to find the min and max price, and then compares the price to these values to get the ids:
select id_prod_ty,
MIN(minPrice) as price_low,
MIN(case when price = minPrice then id end) id_price_low,
MAX(maxprice) as price_high,
MIN(case when price = maxPrice then id end) id_price_high,
id_shop
from (select *,
MIN(IsThisShopPrice) over (partition by id_prod_ty, id_shop) as minPrice,
MIN(IsThisShopPrice) over (partition by id_prod_ty, id_shop) as maxPrice,
from (select p.id_prod_ty, s.id_shop, p.price, p.id,
(p.distance - s.position) as diff,
(case when (p.distance - s.position) = MIN(p.distance - s.position) over (partition by p.id)
then 'Y'
else 'N'
end) as IsThisShop
(case when (p.distance - s.position) = MIN(p.distance - s.position) over (partition by p.id)
then price
end) as IsThisShopPrice
from products p join
shops s
on p.id_prod_ty = s.id_prod_ty
) ps
) ps
group by id_prod_ty, id_shop